What we've decided not to build
Roadmaps get published constantly. Lists of things a company has decided not to build are rarer, which is unfortunate, because they are usually more informative. A feature list tells you what a team found achievable. A refusal list tells you what it values when the two conflict.
This is ours. Each of these has been proposed, seriously, by someone reasonable — usually more than once.
Public follower counts
Follower counts turn a social graph into a scoreboard. Once the number is visible, it becomes a target, and behaviour reorganises around it: posting for reach rather than for people, quiet disappointment at a post that did not travel, and eventually the sense that having a small audience is a failure rather than a choice.
The number is genuinely useful in exactly one context — evaluating a stranger's reach — which is a creator-economy problem rather than a staying-in-touch problem. We are building for the second thing.
The related decision: we do not show how many people viewed a story after it expires, and we do not retain those metrics at all. The value of that number is almost entirely in feeling bad about it later.
An algorithmic "for you" feed of strangers
The recommendation feed is the most commercially successful mechanism in the history of consumer software, and we are not building one. Ranking by predicted engagement reliably surfaces whatever provokes the strongest reaction, and the content that provokes the strongest reaction is not usually the content people are glad to have seen.
We rank a feed of people you have actually chosen. That is a meaningfully smaller and less profitable problem, and it is the one we want to be good at. The trade-offs are set out in A quieter feed.
Read receipts you cannot turn off
Read receipts are useful between close friends and corrosive when mandatory. The corrosion is not really about privacy — it is about obligation. A visible seen converts a message into a debt with a due date, and the debt accrues whether or not the recipient has the capacity to pay it.
Ours are optional and symmetric: turn them off, and you stop seeing everyone else's. We will not ship the version where you can observe without being observed, because that asymmetry is precisely what makes the feature feel unsafe. The same reasoning shapes presence, described in Presence, kept soft.
Streaks
A streak is a mechanic that manufactures obligation and then threatens loss. It is extremely effective at producing daily engagement, and what it produces is not affection for your friends — it is anxiety about a counter. Teenagers in particular describe maintaining streaks with people they no longer especially like, because breaking one feels like a statement.
We would rather people open Seldeo when they want to talk to someone.
Automatic memory resurfacing
"On this day" features are lovely perhaps eighty per cent of the time. The remaining twenty per cent includes a photograph of someone who has died, an ex-partner, a house that was lost, or a pet. Software cannot reliably distinguish these cases, and the cost of getting it wrong is not a mild annoyance — it is ambushing someone with grief on an ordinary Tuesday morning.
Given a feature that is pleasant most of the time and genuinely harmful occasionally, with no reliable way to tell in advance, we default to not shipping it. Given how central memorial accounts are to Seldeo, this one was never close.
Public "last seen" timestamps
A precise timestamp is a surveillance primitive. It can be used to reconstruct someone's day, to check whether they were awake when they claimed to be asleep, and to fuel the specific misery of they were online an hour ago and still have not replied to me. Fuzzy recency conveys the useful part — this message will probably land soon — without the rest.
Ads that follow you around
We will run advertising. We are not building the version that tracks people across other apps and sites to do it, and we do not sell personal data to advertisers. The reasoning, including what we do use for targeting and what we refuse to, is in Our advertising philosophy.
What saying no actually costs
It would be dishonest to present this as costless virtue. Every item above is a growth mechanism that works, and declining all of them has consequences: slower growth, less time spent in the app, weaker viral loops, and a smaller commercial ceiling than a competitor willing to ship them.
We think that is the correct trade, and we would rather state it plainly than pretend the choices were free.
The harder truth is that this list only means something if it survives commercial pressure. Any company can publish principles while small. Publishing them is partly a way of making it awkward for our future selves to quietly reverse course — a list like this is a hostage to fortune, and that is rather the point.
If we ever do ship one of these, we will say so here, and we will explain what changed our minds.